Dr.

Shounak Dasgupta

JU, 6/27/2017 1
 Elements of Digital Communication

PCM , DPCM, DM RZ, NRZ, AMI

Information A/ D Line Digital

Source Converter Encoder Modulator

ASK , FSK, PSK

10100111001

Baseband signal Channel

Information D/ A Digital
Line Decoder
Sink Converter Demodulator

JU, 6/27/2017 2
 A/D Line Digital
Source
Converter Encoder Modulator

JU, 6/27/2017 3


JU, 6/27/2017 4
 Sampling Theorem

fs= 2fm is Min. sampling rate called Nyquist rate.

A band-limited signal of bandwidth fm can be completely recovered from

its samples taken at least at Nyquist rate.
JU, 6/27/2017 5
 Pulse
Modulation

Analog Digital
Pulse Pulse
Modulation Modulation

PCM
PAM , PWM,
DPCM
PPM
DM

 Pulse digital modulation is basically a scheme that converts analog

signal to its corresponding digital form.
 In other words, analog to digital conversion is some times called Pulse
digital modulation.
JU, 6/27/2017 6
Pulse Code Modulation (PCM)
The simplest form of pulse digital modulation.

 The message signal is first sampled (discrete time-continuous

amplitude).
 Then amplitude of each of the sample is rounded off to the nearest
one of a finite set of allowable values, known as quantization levels
(discrete time- discrete amplitude).
 The allowed quantization levels are finally coded to obtain the analog
signal in digital form.

JU, 6/27/2017 7
PCM was invented by the British engineer Alec Reeves in 1937 in France.
It was not until about the middle of 1943 that the Bell Labs people
became aware of the use of PCM binary coding as already proposed by
Alec Reeves.

Application

PCM is the most commonly used technique in digital communications

Used in many applications:
• Telephone systems
• Digital audio recording
• CD laser disks
• voice mail
• digital video etc.

JU, 6/27/2017 8
PCM consists of three steps to digitize an analog signal:
1. Sampling
2. Quantization
3. Binary encoding
Before we sample, we have to filter the signal to limit the maximum
frequency of the signal .Filtering should ensure that we do not distort the
signal, ie remove high frequency components that affect the signal shape.

JU, 6/27/2017 9
Sampling:

 The process of generating pulses of zero width and of amplitude equal

to the instantaneous amplitude of the analog signal.
 The no. of pulses per second is called “sampling rate” which is
determined by the maximum frequency content in the signal (Nyquist
rate).
 The signal is sampled at certain intervals such that each sample is
proportional to amplitude of signal at that instant.
 “Digitization in time” - sampling process results in signal that is discrete
in time but analog in amplitude!

JU, 6/27/2017 10
Quantization of Signals
 The process of converting a discrete-time continuous amplitude signal
into a discrete-time discrete amplitude signal is called quantization.
 This is done by approximating each of the sampled values varying over
a continuous range, to the nearest value from a small set of
predetermined discrete values.
 The predetermined discrete-amplitude levels are called quantization
level.
 The spacing between the two adjacent quantization levels is called a
step-size.
 The quality of a quantizer output depends upon the number of
quantization levels used.

JU, 6/27/2017 11
Quantization Process

JU, 6/27/2017 12
 Types of Uniform Quantizer

 The Mid-tread type is so called because the origin lies in the middle of a tread
of the stair-case like graph. The quantization levels in this type are odd in
number.
 The Mid-Rise type is so called because the origin lies in the middle of a raising
part of the stair-case like graph. The quantization levels in this type are even in
number.
 Both the mid-rise and mid-tread type of uniform quantizer is symmetric about
the origin.
JU, 6/27/2017 13
 Quantization Noise or Error
The difference between the quantized signal and the original signal is
viewed as a noise due to quantization process and called quantization noise
or quantizing error.

The quantized signal and the original signal differ from each another in
random manner. However, the error voltage will lies between - Δ/2 and
Δ/2.

Let a particular sample at the quantizer input lies in the range (k Δ – Δ/2)
to (k Δ+ Δ/2), here k is an integer & k Δ is the corresponding quantizer
output. qe be the error produced
by quantization process. k Δ + Δ/2
qe kΔ
Δ/2 ≥ qe ≥ - Δ/2
k Δ – Δ/2

JU, 6/27/2017 14
JU, 6/27/2017 15
 Signal to Quantization Noise Ratio
For a sinusoid message signal of amplitude A & for L quantization
level
SignalPower
SNR 
NoisePower
For any signal , the mean square value gives the power content.
2
In case of the sinusoid signal, the signal power  A 2
We know for sinusoidal signal, Quantization noise power
2 A2
Pq  
12 3 L2
A2
3 L2
SNR  22 
A 2
2
3L
 3 L2 
( SNR ) dB  10 log 10 
 2
  3
  10 log 10 2  20 log 10 L  1.8  6 n

For n bit encoder: L =2n

For a non sinusoidal signal with normalized destination power & normalized input
message signal (SNR)dB = 4.8 +6n
JU, 6/27/2017 16
Problem:
A PCM system uses a uniform quantizer followed by a 10-bit encoder.
i. Find SNRq for the system
ii. How many bits are required to achieve a SNRq of 40 dB

Ans:

i. (SNR)dB = 1.8+6n = 1.8+60 =61.8 dB

ii. 40 = 1.8 + 6n
n = 7 bits

JU, 6/27/2017 17
Non Uniform Quantization: Companding
o Process of uniform quantization, discussed so far, is not always
desirable.
o For some signals with large dynamic range ( such as voice signal)
where high values occur rarely, variable separation between the
quantization level is advantageous.
o Such variations are taken care of by making use of non-uniform
quantization where the step size increases with amplitude of the
signal.

Voltage Range

Input Signal

JU, 6/27/2017 18
 Why Non-uniform quantization is done :

 Using non-uniform quantization, uniform

quality of service or signal to quantization
noise ratio is maintained through out the
amplitude range in case of a message signal
with large dynamic range. The weak signals are
favoured in non-uniform quantization by
reducing the step size for such amplitudes.
 If uniform quantizer were used, a large number
of steps would have been needed to achieve the
same performance.
 Higher dynamic range with improved coding
efficiency (lesser bits for encoding) is the
advantage of non-uniform quantization over
uniform quantization.
JU, 6/27/2017 19
How Non-uniform quantization is done :
A non uniform quantizer is basically a compressor followed by a uniform
quantizer. The process of non-uniform quantization which is based on
this method is called companding.

In this process the message signal is passed through a compressor and

then is subjected to uniform quantizer.

The compressed samples must be

restored to their value at the receiver
by using a system with a No compression
characteristics complementary to the
compressor. Such system is called
expander.

The compressor and expander

combination is called compander.
JU, 6/27/2017 20
JU, 6/27/2017 21
Compression Techniques
 A large no. of compression laws are present. Among them, two laws e.g,
μ-law and A-law meets the desirable standard by CCIT (International
Telegraph Consultative Committee).
 Telephones in US, Canada and Japan use μ -law Companding. (μ =255)
 A-Law is used elsewhere to compress digital telephone signals.

JU, 6/27/2017 22
 μ -law
μ –law is defined as Vi and Vo are normalized input & output

 Compressed output is linear for low input

& logarithmic for high Input volt.
 μ = 0 corresponds to no compression.
 μ = 255 is a typical value used for
telephony system in US, Canada.
JU, 6/27/2017 23
 A -law

A-law can be defined as

 Uniform quantization corresponds

to A=1
 A = 87.56 is a typical value used in
practice.

JU, 6/27/2017 24
The main advantages and drawbacks of the two compression
algorithms are:

 A-law provides a greater dynamic range than µ-law. Dynamic range is

the ratio in decibels between strongest and weakest signals.
 µ-law provides better signal-to-distortion performance for low level
signals than A-law leading to higher signal fidelity at low levels.
 A-law requires 13-bits for a uniform PCM equivalent while µ-law
requires 14-bits for the same uniform PCM equivalent.
 Uniform PCM equivalent is the number of bits necessary to represent
the compressed signal using uniform sampling intervals.
 An international connection will always use the A-law. For example, a
trans-Atlantic link connecting a µ-law country (US or Canada) with an
A-law country (any country in Europe) by definition will use the A-law.
 The µ-law to A-law conversion is the responsibility of the µ-law
country(the North American country).

JU, 6/27/2017 25
Coding:
The process sampling followed by quantization converts an analog signal
into a discrete time signal whose amplitudes are limited to a set of discrete
values.
The discrete set of sample values are then represented by symbols through
binary coding.
In binary coding, each code word consists of n no of fixed bits to represent
2n no of quantization level. If M be the no quantization levels then,
M ≤ 2n

Digital O/P: 100 111 110 011

JU, 6/27/2017 26
Bit rate & Bandwidth in PCM system:
• If B Hz be the maximum frequency present in the message signal, then the
Nyquist rate of sampling fs = 2B Hz. It means 2B samples are taken per sec.
• Now, let n be the no. of bits require to encode individual symbols, then total no.
of bits is used per sec = n fs = 2nB bps
• This is the minimum bit rate require to send the message without error. Many a
time , bandwidth requirement is also specified by this value.
• The channel capacity of the channel should be at least equal to this bit rate in
order to transmit the data.

Example : An audio signal consists of a sinusoidal term given by

f(t) = 3 cos ( 1000 πt)
The signal is sampled at trice the nyquist rate & fed to a 64-level quantizer
followed by a binary encoder. Find the minimum data rate to be supported by
the channel to send the data.
Ans: Frequency of the signal B = 500 Hz & Nyquist rate fNq= 2B = 1000 Hz
Sampling frequency fs = 3 x fNq = 3000 Hz, i.e, 3000 samples per sec.
To encode 64 levels , we require 6 bits each for the samples ( 64 = 26 )
Thus the required bit rate = n fs = 6 x 3000 =18000 bps = 18 kbps
27
JU, 6/27/2017
Bandwidth in Hz:
For a n-bit PCM with the sampling rate fs = 2B Hz, the sampling period :
1
Ts 
fs
Let us consider N different channels are being transmitted in this PCM system.
Thus total no. of bits per sampling period Nt= nN bits
Then the individual bit duration
Ts 1 1
Tb    sec
Nt nNf s 2nNB

We usually represent bits in PCM by rectangular

pulses of width Tb and the practical bandwidth
of such signal is given by
1
BW   nNf s  2nNB Hz
Tb
Bandwidth in Hz is numerically equal to the required bit rate.
JU, 6/27/2017 28
Example:

24 telephone channels, each band limited to 3 KHz, are to be TDM by

using PCM. Calculate the bandwidth of the PCM system for 128
quantization levels and 8 KHz sampling frequency.

Ans:
Given, N = 24, n = log2 128 = 7 bits, fs = 8000 Hz

BW = 7 x 24 x 8000 Hz = 1.344 MHz

JU, 6/27/2017 29
Trade off between SNR and bandwidth

For sinusoidal signal, Signal to Quantization noise power ratio

Let n be the no. of bits require to encode individual symbols, the

minimum bit rate require to send the message without error = n fs = 2nB
bps , if B Hz be the maximum frequency present in the message signal.

 In order to increase SNR, the step size Δ is to be reduced by

consequently increasing no. of levels L which in turn will increase
no of bits per sample n. It results in increase in required bit rate.
 Thus trade off between SNR & bandwidth is to be made.

JU, 6/27/2017 30
Standard Requirements of Different PCM systems:

 Voice: 4k Hz x 2 = 8000 samples/s x 8 bits (256 levels) = 64 kbits/s

 Video: 4M Hz x 2 = 8,000,000 samples/s x 8 bits = 64 Mb/s
 Cable TV supports 300 video channels = 19.2 Gigabits/s

JU, 6/27/2017 31
Consider a signal f(t) = 20 cos (2000t). The signal is sampled at
Nyquist rate & fed to a 7 bit encoder after uniform quantization.

a. What is the change in SNR if encoded bit is increased by 1. What

is the change bitrate of the system.
b. What will be the bit rate if we double the sampling rate.

Ans:
Given, fm = 1000 Hz, fs = 2000 Hz & n=7

a. SNRq = 1.8 + 6n ; SNRq7 = 43.8 dB & SNRq8 = 49.8 ;

ΔSNRq = 6 dB
Br = n fs ; Br7 = 7*2000 = 14 Kbps , Br8 = 8*2000 = 16 Kbps
Δ Br = 2 Kbps .
b. fs‘ =2 fs , Br = 2*7*2000 = 28 Kbps

JU, 6/27/2017 32
Advantages of PCM system:

 Uniform Transmission Quality

 Compatibility of different classes of Traffic in the Network
 Low Manufacturing Cost
 Good Performance Over Very poor Transmission Paths

JU, 6/27/2017 33
Disadvantages of PCM system:

χ Large Bandwidth required for Transmission

χ Noise and crosstalk leaves low but rises attenuation

JU, 6/27/2017 34